🔍 Read the full analysis: MiniMax M3.1 Flash Trails The Leaderboard But Leads Our Newsroom Test on ThorstenMeyerAI.com
Prime made for students and young adults
- Fast, free delivery for dorm and study essentials
- Prime Video and Amazon Music included
- Member-only deals
TL;DR
MiniMax released M3.1 Flash on October 19, 2026, at $0.15 per million input tokens and $0.60 per million output tokens, while promising to release its weights on October 23. An independent index placed it behind several open-weight rivals, but the supplied source does not include the newsroom test’s scores, so its headline claim that the model leads that test cannot be independently assessed here.
MiniMax released M3.1 Flash on October 19, pricing it at $0.15 per million input tokens and $0.60 per million output tokens. The company presents the model as a fast, low-cost option for agent tasks, but an independent index scored it below several open-weight competitors; the source describing a newsroom test does not provide the test’s results.
MiniMax made M3.1 Flash available through its API, M Plan subscription, MiniMax Code and OpenRouter. OpenRouter offered access free for the first 72 hours. The listed price for cached input is $0.015 per million tokens. For comparison, OpenRouter lists the earlier M3 at $0.30 per million input tokens and $1.20 per million output tokens, making the new model half the price per token on those measures.
MiniMax says M3.1 Flash can generate more than 200 tokens per second in everyday use, with a median above 170, and describes it as nearly three times faster than M3. Those are company claims, not the result of the independent measurement cited in the source. Artificial Analysis measured 120 tokens per second and 1.08 seconds to the first token. The release lists a one-million-token context window; reasoning effort can be adjusted. MiniMax has not disclosed an official parameter count.
The weights are not yet available. MiniMax said they would arrive on October 23, but the source provides no confirmation that the release occurred. Until the weights are published, Artificial Analysis classifies M3.1 Flash as proprietary. MiniMax also says the model costs about 70% less per agent task than M3, a separate company claim that depends on task-level usage, not just token prices.
Infographic 01 · MiniMax M3.1 Flash
A step behind on the index. Ahead of every open model in our newsroom.
An independent intelligence index puts MiniMax M3.1 Flash a step behind the open-weight leaders. Writing real news from real sources, it beat all of them.
Independent index
Artificial Analysis Intelligence Index, 10 evaluations. Bar length on a 0–50 scale.
Artificial Analysis lists M3.1 Flash as proprietary until its weights are released.
Our newsroom test
Mean score, English articles, 1–5 by GPT-6.1 Sol and Claude Opus 5.5 with the source in context. Bar length on a 0–5 scale.
In German, M3.1 Flash (3.36) also leads every open model, but GPT-6 Luna is clearly ahead (3.62).
Sources: Artificial Analysis, MiniMax-M3.1-Flash and open-weights models (20 Oct 2026); Thorsten Meyer AI newsroom benchmark, 12 sources × English + German, run 20 Oct 2026.
Powered by Thorsten Meyer AI
Price, Speed and Real-World Use
The launch puts a lower-priced model into a crowded field where token cost alone does not determine value. Artificial Analysis reported that M3.1 Flash used 180 million tokens to complete its index evaluation, compared with a 99-million-token median. That means a cheaper rate per token may not translate directly into lower costs for a given workload if the model uses more tokens to finish it.
The gap between MiniMax’s speed claim and Artificial Analysis’s measured speed also matters for teams choosing models for time-sensitive agent work. The index measured M3.1 Flash at 120 tokens per second, below the company’s stated everyday-use figure of more than 200 and below DeepSeek V4.1 Flash 1017’s 211 tokens per second in the same comparison. Different test conditions may affect results, and the source does not provide enough detail to reconcile them.
For newsroom and publishing use, benchmark rankings do not answer whether a model preserves source facts or invents details. The described 24-article evaluation was designed to test those qualities across English and German articles, using two judges. But without the scores or comparative results, readers cannot tell whether M3.1 Flash actually led that test or how it performed against other models.
As an affiliate, we earn on qualifying purchases.
From Space Bunny to M3.1
Before the public launch, the model appeared anonymously on OpenRouter as Space Bunny Alpha. According to the source, it went live on September 23, led OpenRouter’s usage ranking for 10 days and processed 56.7 trillion tokens on the platform before being removed on October 5. MiniMax says its total use across services exceeded 200 trillion tokens. These usage figures are attributed to MiniMax or the source account; they do not by themselves establish model quality.
The anonymous period raises a data-handling question for users who sent prompts to the endpoint. The source says the provider could retain prompts during that period but was not allowed to use them for training. MiniMax has not explained what happened to that stored data. The source also says no license had been announced for the forthcoming weights, and it excludes an unconfirmed parameter-count estimate reported elsewhere.
MiniMax’s published benchmark charts compare M3.1 Flash with a selection of rivals, including DeepSeek V4.1 Flash 1017, Kimi K3, GLM-5.3, GPT-5.6 Sol and Claude Opus 5. In those company-reported results, M3.1 Flash scored 62.2 on Automation-Bench, ahead of DeepSeek’s 60.5, while DeepSeek scored higher on DeepSWE and Terminal-Bench. The charts do not report every model for every test, and the results have not been independently reproduced, according to the source.
““nearly three times faster” than M3”
— MiniMax
As an affiliate, we earn on qualifying purchases.
Missing Scores and Release Details
The source’s headline says M3.1 Flash leads a newsroom test, but the supplied material ends before the results table. It gives the test design—24 articles in English and German, judged by GPT-6.1 Sol and Claude Opus 5.5—but no scores, model comparisons or judges’ findings. The claimed newsroom lead therefore cannot be verified from the available information.
Other open questions include whether the promised weights appeared on October 23, what license governs them, and whether MiniMax will disclose the model’s parameter count. The source also does not say how long prompts from the anonymous OpenRouter period were retained or what happened to them. MiniMax’s benchmark results and security figures are described as self-reported and not independently reproduced.
Comparisons across speed and cost also need care: the source does not provide enough testing-method detail to explain the difference between MiniMax’s speed claims and Artificial Analysis’s measurement, or to establish whether the token-use finding applies to typical customer tasks.
As an affiliate, we earn on qualifying purchases.
Weights and Test Results Ahead
The next stated milestone is MiniMax’s promised October 23 weights release. If published, the weights and their license could allow outside researchers to assess the model directly; the source does not confirm either the release or its terms. MiniMax could also clarify its parameter count and the handling of prompts submitted during the Space Bunny Alpha period.
For readers evaluating the newsroom claim, the missing score table is the key outstanding item. Publishing the English and German scores, comparison models and judging method would show whether M3.1 Flash led and how much the result depended on language or evaluation criteria. Until those figures are available, the confirmed picture is a low-priced release, a mixed independent benchmark showing and a newsroom test whose reported outcome is not included in the supplied source.
As an affiliate, we earn on qualifying purchases.
Key Questions
What is MiniMax M3.1 Flash?
M3.1 Flash is a model released by MiniMax on October 19, 2026. The company markets it for fast, lower-cost agent work and made it available through its API, subscription, coding product and OpenRouter.
How much does M3.1 Flash cost?
The listed price is $0.15 per million input tokens, $0.60 per million output tokens and $0.015 per million cached input tokens. MiniMax separately claims it costs about 70% less per agent task than M3; the source does not provide an independent verification of that claim.
Did M3.1 Flash lead the newsroom test?
The source headline says it did, and the article describes a test of 24 English- and German-language news articles. However, the supplied material does not include the results table or scores, so the claimed lead cannot be checked from the information provided.
How did it score against open-weight models?
Artificial Analysis gave M3.1 Flash an Intelligence Index score of 43, below several models in the cited comparison, including Xiaomi MiMo-V2.6-Pro at 46 and DeepSeek V4.1 Flash 1017 at 45. It measured M3.1 Flash at 120 tokens per second.
Are the model weights available?
Not according to the source, which says MiniMax promised the weights for October 23, 2026. It does not confirm whether they were released or specify a license.
Source: ThorstenMeyerAI.com
Halloween Picks
halloween
As an affiliate, we earn on qualifying purchases.
